Hey folks,

I'm currently working on a "semi-binaural" setup for some field recordings and would like to share my present progress. I would be glad if you had some ideas how to improve the rig and provide it some more immersion

My gear consists of two Primo EM172 capsules connected via XLR (P24 pulled down to ~10V) to a Tascam DR-60D recorder. The capsules are placed right at the canal entrance of two silicon ears. Both ears are attached to a small plastic base (just an old mic case I had laying around) via some elastic bands.
I have attached a picture below. Nothing too exciting, but does the job for the moment.

The reason I initially called this construction "semi-binaural" was that I do not use a head model as a base which evidently spoils the HRTF.
However, I am wondering whether you have some experiences with (semi-)binaural setups that are build for a mobile and rather compact purpose.
The goal is to capture as immersive as possible (not necessarily scientificly accurate) ambient sounds with a construction that is - how should I express it - not making me look like I'm coming from another dimension

I know that utilizing a head model is essential for retaining the physical propertiers that ensure a lifelike sound. Nevertheless, there are a lot of builds that consist of a minimalistic base that appears to only fulfil a static role (e.g. 3Dio) and still sound pretty immersive (at least to my ears).

Are there any experiences with DIY solutions of such kind?

Some ideas I had so far:

1.) Adding some plates right behind both ears to improve (at least) the frontal plane. Just like a duo jecklin disc.

2.) Adding ear canals.
I have already ordered some 7mm/10mm (inner/outer diameter) plastic tube and wanted to give it a try. However, this technique appears to be treated kinda controversially since it potentially doubles the sound path and requires havier EQ in post processing.

3.) Using a black cylindric base to connect both ears with a distance of maybe 20cm. The inside would be filled with absorbing material such as foam or even mineral wool. Compared to a fullsized head this construction wouldn't be that bulky and could possibly be carried around more inconspicuously. The (skin-colored) ears themselves could be cloaked by some black "dead cat" fur or foam.

4.) Maybe sticking the whole thing to the backside of my backpack instead of carrying it around in the hand for mobile use.
But I'm generally curious about the impact of my body plus the backpack (which obviously form acoustically resistances) on the spatial perception.

Any impressions on those subjects?
I would be delighted if you could give some tips and help me improving the setup

By the way, here are two initial test recordings I made with the structure shown in the picture below:




Best Regards,
Pepi

No impressions? :'(

Hello Pepi.
I would definitely encourage you to experiment with the plastic tube ear canals (get capsules capable to handle high SPL!). The equalization is not nearly as difficult as many people seem to think. As a matter of fact, you get that EQ with EXACTLY the same process that you follow to equalize pinnae only DIY binaural microphones (you REALLY should equalize those as well, even though some people get away without doing it, when the recorded material is not music).

I would also get a dodgeball and stick all the apparatus inside (with pinnae sticking out, of course). Head shadow effect is a very important part of HRTF, and the 3Dio approach, while creatively interesting, is not proper binaural by any stretch of imagination.

Hey sax,

thank you for your response!

It's been quite a while since I worked with the "3Dio" kind of approach. And overall, I wasn't really that impressed with the results of that setup.

In the meantime, I went further and bought a dummy head which also comes with the upper part of a torso. I haven't attached the ears yet since I'm still reflecting about the size of the head and the placement of the ears / capsules. The dummy's head girth is about 54cm with a bizygomatic breadth of approximately 12,5cm, so it's probably more in the average size of a 10-14 year old. Since the head is a bit smaller than the reference of a fully-grown person, I'm afraid that there will be less shadowing and separation between both channels. The smaller the head, the shorter the distance the sound has to travel around the head from one ear to the other. Thus, there could be a weaker sound stage and maybe even worse localization.

However, I could possibly improve that by extending the sound path inside of the head. The simulation of the ear canals by attaching some tubes could increase the time difference from one capsule to the other and thus compensate for the smaller head size. However, I should probably make sure that there is no greater leakage from capsule to capsule inside of the head then (-> filling the inside with damping material).

Looks like there is still some trial and error to do.

Do you have any suggestions on that relations?

And what is the best way to compensate for the frequency deviations of the whole setup? Play some pink noise in front of each ear and try to equalize it flat in the DAW?

Regards,
Pepi

I went down a similar path, and this is where I got stuck.

Flat response directly in front, at +/- 30 degrees, averaged flat over a range of angles..?

Never quite got there, although I do have the prototype around here somewhere. I was just using a pair of ECM8000s, since I happened to have them around.

Following with interest.

Sax, great to see you here.

Chris

Hey Chris,

in theory you should measure the response on the 0° axis with a flat full range loudspeaker (point source) I guess. Measuring apart from the direct 0° field would already colour the audio and thus distort the baseline of the EQ curve.

I'm planning to build a small one-way speaker for mic measuring purposes soon. Hope it will be mostly flat when I stuff the speaker into a small MDF case. However, should be easy to compensate by EQ, when capturing its response with a measuring mic

Regards,
Pepi

Hey Pepi,

The trick there is to make a calibration file that includes the speaker.


Okay, so if we measure directly in front of the recording head, what happens when we move to the traditional "stereo" equilateral triangle?
Is it best to measure from two sources simultaneously?

You can see how I got tied up in all this. I once went as far as to make around 20 measurements (for one ear), all in the horizontal plane going around 360 degrees, to average them.
That didn't work, either.

Chris

Hello. Sorry, I missed your replies..

I would start by saying that there is non substitute for correct head dimensions.
Extending the ear canals would do more damage than good, since the time it takes the sound to get to the canal opening is the delta T encoded in the recording. After that, at the end of the ear canals, the length of them being equal, the delta T is maintained constant.
So unfortunately one has to find a correct head, which is not as easy at it sounds. I tried many before finding a good one, so good hunting!

As far as the EQ procedure, see the first post of this thread I started (A better binaural microphone), where I explain the EQ procedure and why I insist so much on preserving the geometric boundary conditions, if one is interested in accuracy.
In short, the EQ tries to 'undo' the playback part of the audio chain, so it's found by putting the mic in the sweet spot of a traditional stereo setup (I actually use about +/- 45deg placement, because it maximizes cross talk cancellation).

The idea of building a single driver to find the EQ is valid. I'd like to know how it goes, since I'm working on something similar.

Great to see you here too, Chris.

Hey there,

thanks for sharing your impressions and experiences again!

Actually I have found another model a few weeks ago which is bit larger. I'm still thinking about the way to design this dummy head. Here are a few of my considerations and uncertainties so far:

1) Blocked or or open ear canal?
The blocked version is easier to handle, of course. However, the reproduction of the canal will probably allow more accurate results with better timbre.

I'm still researching how to mold with silicone. Might pour a replica of my ear canals and stick them to my ear models which I bought from a an accupuncture store quite a while ago. Maybe I will also pour some molds of my own outer ears (pinnas). But I still have to figure out how to do so, with pleasing results and proper safety precautions (fully blocking the ear canal etc.).

2) Reinforcing the head model.
The dummy head is made of cheap plastic which introduces unpleasant resonances. It is not the thinnest material, at least you can't compress it with your finger. But still on the lighter side.

I might reinforce it with some sheet of bitumen which improves the insulation and reduces vibrations. Or maybe some layers of plaster to approximately reproduce the inteference with the human skull?

3) Damping the inside.
Got a few bags of shredded foam left. Filling all the volume of the head with slight compression of the material should probably avoid most of the internal (in-head) crossfeed, only leaving the canals of the ears as primary sound paths.


Those are the main constructional aspects I'm thinking about at the moment.
Any further ideas or contradictions (misconceptions) so far?

- - - - -

Thank you the info regarding the EQ by the way! I will look into it as soon as the editing phase comes closer.

The measuring speaker is still in the planning stage, as well.
Since good coaxial speakers are really pricey and hard to get, I might go for some 4.0" full range drivers by Tang Band (like the W4-655). I have heard that such drivers can play down to 50 Hz with just 6-7l of inner volume with some builds, without much need of equaliation and linearization.

More to come.
I will keep you up to date

Regards

This is all a lot of fun as a hobby. But it will be just about impossible to match the performance of the KU-100 without a significant capital investment.

I've gone through all this and at the end I had to model my 3D printed pinnae+canals.
Trying to attach a canal replica to a separate ear model is very difficult and, more than that, pointless. One goes through all the trouble of getting a canal replica (and it's A LOT of it) only to mess the geometry at the junction between the canal and the pinna.
You might as well use simple clear plastic tubes. The junction isn't smooth and the geometry approximate, but I still got very good results with something as simple as that.
Make sure you get the EQ right when using ear canals. It is not as forgiving a rig as with just pinnae.

You can kill these two birds with the same stone. Get a block of foam and put it inside the head, making sure it applies pressure to the walls. The shredded pieces are going to be difficult to make stay in there and maintain pressure on the inside, but if you manage to do it more power to you.
One doesn't need to replicate the weight and consistency of the human head. The direct sound impacting the eardrums is much higher in power than these secondary effects. If you're worrying about these things, get the canals and EQ right first. Get your priorities in logical order.

This is incorrect. I have achieved at least same quality performance as the KU100 recordings I could find out there with a DIY microphone. It used clear plastic tube ear canal replicas, so not even the anatomically accurate canals.
Again, getting the EQ right is paramount.
Everything said and done I might have been $200-300 into it (but also, I must admit, a lot of man hours).

As an interested party who has lurked around the binaural concept without committing (yet) to the significant investment in time or in money required to achieve high quality recordings, I would be very interested to compare some examples of commercial and DIY dummy head recordings. (I realize this won't be ABX, as it is unlikely that someone who owns a KU 100 would be building their own dummy head rig and vice-versa). Perhaps you could put up a sample of "the best you have" of music with a full dynamic and frequency range and someone who owns a KU 100 could post something as well. Class, any takers?

I would gladly record the same concert side by side with a KU100 or any other binaural microphone and compare in a double blind test, if somebody is in the DMV area.
For now, this is an excerpt from the following article. The mentioned recording was done with clear plastic tubes as ear canals.
Flying Blind: Another Visit to the Magnepan Skunkworks - The Audio Beat - www.TheAudioBeat.com

Wendell played a couple of demo cuts he had played for me before, and then he played some music I'd never heard: orchestra and a pair of vocalists recorded in a church. The recording quality was stunning -- more natural and powerfully real than I've ever heard while at Magnepan. I asked Wendell about the recording, and he told me that it was the work of a little-known professional sound engineer from Washington, DC, Marco Sabatini. He was an admirer of Magnepan speakers and had contacted the company, eventually offering some of his recordings for demonstration purposes. I should say so! This cut, of primarily vocal music, had a massive sense of space, and the placement of the voices was so precise that, while I couldn't see a thing, I felt like I was watching the performers in front of me. As often as I've heard music in this room, in the dark, this might have been the best sound ever. At the very least, it raised the issue.

My initial plan was to get an ear mold made by an audiologist in the first step and then moulding a copy of the outer ear at home. I would then stick the mold of the ear canal into the replica of the pinna and pour the negative ear canal around it. However, I don't know how to keep those parts together aftwards, since the first piece (pinna) has already dried before (afaik only fluid silicones will fuse together). Maybe some 3M silicone tape can help holding the two surfaces together. But I don't know if it is worth the risk and effort after all.

Reproducing the whole thing as 3D print wouldn't be an option since I don't have the equipment and know-how for scanning and printing.

Going with my "generic" pair of silicon ears and simply trying to extend them with some plastic tube would be easier for sure. But I have my doubts that this setup will deliver pleasing results in the long run.

Stability isn't much of a problem. But I doubt that the plastic has the proper acoustical properties for approximately replicating a human head. Thus, I wanted to at least give it some more mass and thickness.

As for the absorption: I could stuff the shredded foam into a nylon bag. One for each half of the assembly. That would prevent the material from spilling out when opening the unit for adjustments.

Good point!


Regards

I went through the same process that you described, trying to pour a negative after sticking the solid canal replica in a hole in a normal pinna replica. In the end I just gave up. It doesn't stick, it looks like crap, and it's not smooth. If you really want accurate ear canals you can buy mine. I sell them on Shapeways.

I maintain that getting the material similar to the real human head is at least one order of magnitude less important than getting the geometry right by employing accurate ear canals. However, just to get your feet wet, I would suggest trying clear plastic tubes at first. You can get quite good results with those.

The plastic bag is a good idea, I think. Just make it tight so it doesn't crackle like plastic usually does.

Thanks for the article; an interesting read and high praise indeed. Could you post a bit of the recording mentioned in the article? Or, if not possible, could you post some other clip amply demonstrating the "binaural with ear canals" sound?

To me, in some respects, binaural is still an undeveloped frontier while ambisonics has been getting most of the attention of late...

Unfortunately, I haven't worked much on recording in the last 2 years. There is no money and no interest in good audio recordings in my area.
But that's ok. I'm off to new and hopefully more exciting things.

This is one of the latest recordings I made. There is a little added EQ because the client wanted it to sound warmer. It took away some of the clarity, in my opinion. Unfortunately, it's youtube sound quality, but it should still render the idea.

YouTube

In response to the OP, a great mobile solution is to use your personal pre-fabricated human head with small lavalier omnis attached to glasses on either side of it. For best results I like to keep my head attached to my neck and shoulders at all times. I did a few church and street recordings like this with nice results back in my early days as a recordist- very discrete. If you don't own your own human head, contact your local mob boss but, fair warning, that's a pricier option than the neumann head and far less discrete.

If we want to stay in the quasi-binaural realm, this is a very simple approximation of the concept I made a while ago. My wife named it Marvin, from the martian :-)
It sounds very good without much of an EQ.

250 USD + shipping to Germany + import sales tax + customs duties ...
Ouch...

Thanks for the advice. I might give it a try.

Not plastic, nylon. So there will be no crackling

Looks really nice!
May I ask which mics / capsules you are working with? And how are they oriented (angle and distance)?

I haven't done a lot of research about spherical microphones (or "Kugelflächenmikrofon" as we call it), yet.
Would you say there are any significant advantages or disadvantages compared with DIY dummy heads? There will be less tonal inteferences due to the lack of an ear (canal), of course. But what about the localization and spaciousness?

Thats one of the easiest and inconspicuous attempts, of course!
But sadfully, they won't include the effects of the ear canal.


Regards


Edit:
I have found a basic paper about "Trennkörper-Mikrofonsysteme" ("separating body microphone systems") comparing various builds:
https://curdt.home.hdm-stuttgart.de/...fonsysteme.pdf

In short:

The KFM (spherical mic) appears to be an interesting alternative.
I will keep that one in mind for future projects

It all depends on how much you value your time. I can tell you that following through with what you have in mind is going to require A LOT of time, only to get to a result that is not anatomically accurate anyway.

$250 is a decent sum of money (although peanuts when you consider we are talking about audio stuff), so I guess my initial suggestion to try clear plastic tubes, at least at first, is still my best advice to you.